**[OpenAI Launches New Voice Architecture GPT-Live: Supports Simultaneous Listening and Speaking, Smooth Background Processing]** According to monitoring by Beating, OpenAI has announced the rollout of the third-generation voice model and architecture, GPT-Live, to global ChatGPT users. Free-tier users can access GPT-Live-1 mini, while paid users will default to GPT-Live-1. This system separates complex reasoning from real-time conversations, allowing it to listen while users speak, eliminating the need to take turns like a walkie-talkie. GPT-Live adopts a full-duplex architecture, enabling simultaneous two-way listening and speaking. During conversations, it uses filler words like "uh-huh" and "right" to indicate active listening, supporting quick back-and-forth interactions or remaining silent when users need time to think. For tasks requiring searches or deep reasoning, GPT-Live delegates these to GPT-5.5 in the background while maintaining the conversation with the user. Additionally, during interactions, it can display visual cards for weather, stock market updates, and more to aid understanding. [Original Link]